def partition_volumes(): volume_indices = np.array(range(NUM_VOLUMES - NUM_OFFSET_VOLUMES)) clean_slice_mask = ss.SceneSlicer().get_clean_slice_mask() volume_indices = volume_indices[clean_slice_mask] np.random.shuffle(volume_indices) num_train = int(TRAINING_PERCENT*len(volume_indices)) train_indices = volume_indices[:num_train] test_indices = volume_indices[num_train:] return train_indices, test_indices
def setup_test():
with open('test_scenes.csv', 'w') as csvfile:
scenewriter = csv.writer(csvfile, delimiter=',', quotechar='"')
scenewriter.writerow([17.0, 'SAVANNAH', 'DAY', 'EXT'])
scenewriter.writerow([40.0, 'DOCTORS OFFICE', 'NIGHT', 'INT'])
scenewriter.writerow([61.0, 'GUMP', 'DAY', 'EXT'])
scenewriter.writerow([82.0, 'GUMP', 'DAY', 'EXT'])
scenewriter.writerow([91.0, 'GUMP', 'NIGHT', 'INT'])
return scene_slicer.SceneSlicer('test_scenes.csv')
def setup_test():
data = np.zeros((2, 2, 2, 50))
img = nib.Nifti1Image(data, affine=np.diag([1, 1, 1, 1]))
nib.save(img, 'test_data.nii')
with open('test_scenes.csv', 'w') as csvfile:
scenewriter = csv.writer(csvfile, delimiter=',', quotechar='"')
scenewriter.writerow([17.0, 'SAVANNAH', 'DAY', 'EXT'])
scenewriter.writerow([40.0, 'DOCTORS OFFICE', 'NIGHT', 'INT'])
scenewriter.writerow([61.0, 'GUMP', 'DAY', 'EXT'])
scenewriter.writerow([82.0, 'GUMP', 'DAY', 'EXT'])
scenewriter.writerow([91.0, 'GUMP', 'NIGHT', 'INT'])
return scene_slicer.SceneSlicer('test_data.nii', 'test_scenes.csv')
def get_design_matrix():
"""
Returns
-------
Design matrix with 5 columns, including the 3 columns of interest,
the linear drift column, and the column of ones
"""
data = nib.load(data_path)
n_trs = data.shape[-1]
X = np.ones((n_trs, 3))
ss = ssm.SceneSlicer('test_data.nii', 'scenes.csv')
day_night, int_ext = ss.get_scene_slices()
X[:, 1] = np.linspace(-1, 1, n_trs)
X[:, 2] = day_night
return X
def get_design_matrix(self):
"""
Returns
-------
Design matrix with 3 columns, including the column of interest,
the linear drift column, and the column of ones
"""
if self.design is None:
scene_path = dp.get_scene_csv()
ss = ssm.SceneSlicer(scene_path)
interest_col = ss.get_scene_slices()[self.interest_col_ind]
n_trs = self.data.shape[-1]
design = np.ones((n_trs, 3))
design[:, 1] = np.linspace(-1, 1, n_trs)
design[:, 2] = interest_col[NUM_OFFSET_VOLUMES:NUM_OFFSET_VOLUMES +
n_trs]
self.design = design
return self.design
def partition_volumes():
"""
Partitions the volumes from the SceneSlicer object randomly to generate
training and testing data
Parameters
----------
None
Returns
-------
train_indices : array
test_indices : array
"""
volume_indices = np.array(range(NUM_VOLUMES - NUM_OFFSET_VOLUMES))
clean_slice_mask = ss.SceneSlicer().get_clean_slice_mask()
volume_indices = volume_indices[clean_slice_mask]
np.random.shuffle(volume_indices)
num_train = int(TRAINING_PERCENT * len(volume_indices))
train_indices = volume_indices[:num_train]
test_indices = volume_indices[num_train:]
return train_indices, test_indices
def __init__(self, data_file):
self.X = np.load(data_file).T[NUM_OFFSET_VOLUMES:, :]
ss = ssm.SceneSlicer()
self.y = np.array(ss.get_scene_slices()[0][NUM_OFFSET_VOLUMES:])